Light-Responsive Monobodies for Dynamic Control of Customizable Protein Binding 16 17 César Carrasco-López1,#, Evan M

bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Light-responsive monobodies for dynamic control of customizable protein binding 16 17 César Carrasco-López1,#, Evan M. Zhao1,#, Agnieszka A. Gil2,#, Nathan Alam1, Jared E. 18 Toettcher2,* and José L. Avalos1,3,* 19 20 1 Department of Chemical and Biological Engineering 21 Princeton University, Princeton NJ 08544 22 23 2 Department of Molecular Biology 24 Princeton University, Princeton NJ 08544 25 26 3 Andlinger Center for Energy and the Environment 27 Princeton University, Princeton NJ 08544 28 29 # These authors contributed to this work equally 30 31 * Co-corresponding Authors 32 [email protected]; [email protected] 33 34

1 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

35 ABSTRACT 36 37 Customizable, high affinity protein-protein interactions, such as those mediated by antibodies 38 and antibody-like molecules, are invaluable to basic and applied research and have become 39 pillars for modern therapeutics. The ability to reversibly control the binding activity of these 40 proteins to their targets on demand would significantly expand their applications in 41 biotechnology, medicine, and research. Here we present, as proof-of-principle, a light-controlled 42 monobody (OptoMB) that works in vitro and in vivo, whose affinity for its SH2-domain target 43 exhibits a 300-fold shift in binding affinity upon illumination. We demonstrate that our aSH2- 44 OptoMB can be used to purify SH2-tagged proteins directly from crude E. coli extract, achieving 45 99.8% purity and over 40% yield in a single purification step. This OptoMB belongs to a new 46 class of light-sensitive protein binders we call OptoBinders (OptoBNDRs) which, by virtue of 47 their ability to be designed to bind any protein of interest, have the potential to find new 48 powerful applications as light-switchable binders of untagged proteins with high affinity and 49 selectivity, and with the temporal and spatial precision afforded by light. 50

2 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

51 Introduction 52 The high binding affinity and selectivity of antibodies and antibody-like proteins, such as 53 monobodies, nanobodies, affibodies, anticalins and DARPins, have made them central tools of 54 modern medicine, biotechnology, and basic and applied research1–7. Because of their ability to be 55 raised, selected, or designed to bind practically any protein of interest6,8,9, these protein binders 56 have found multiple applications in diagnostics3,10,11, therapeutics4,12,13 and biologics 57 manufacturing14–16. As such, they have revolutionized the way we treat disease13,17–19, purify 58 proteins20–23, and study biological phenomena9,24–27. The enormous impact that high affinity 59 protein binders have had in medicine, biotechnology, and research stems from their usually 60 irreversible binding to very specific targets. However, their repertoire of applications could still 61 be greatly expanded if they were engineered with optogenetic control over their binding 62 affinities, such that they could bind their targets instantly and reversibly depending on light 63 conditions, while maintaining their characteristically high affinity and selectivity. 64 65 Protein binders can be developed from a variety of scaffolds including based on immunoglobulin 66 domains28,29 or other small single-domain proteins2,17. Monobodies belong to a class of high 67 affinity protein binders that have non-immunoglobulin scaffolds30,31. They are derived from the 68 10th domain of human fibronectin, engineered to structurally and functionally resemble 69 nanobodies28,30,31. As antibody mimetics with human origins, their use as biologic drugs is 70 expected to substantially reduce unwanted immune responses17,32. Their high affinity, specificity 71 and straightforward expression in multiple cell types also make them a versatile tool for 72 research6,31. The small size and relative stability of monobodies (less than 100 amino acids), as 73 well as their lack of disulfide bridges, robust activity inside33 and outside of cells34, and their 74 ability to be selected to bind countless different proteins with high affinity and selectivity, make 75 them ideal candidates to develop light-switchable protein binders by fusing them to light- 76 responsive proteins. 77 78 A protein domain widely used for optogenetic tool development is the second light, oxygen, and 79 voltage (LOV) domain from the oat, Avena sativa, photosensor Phototrophin 1, called the 80 AsLOV2 domain35. This domain elicits its light response through a large conformational change 81 (Fig. 1a) triggered by the formation of a covalent bond between a photoexcited flavin

3 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

82 chromophore, FMN, and a conserved cysteine36,37. The C-terminal helix, Ja, of AsLOV2 is 83 packed against its core domain in the dark. Upon blue light stimulation (optimally 447nm) the 84 covalent bond between FMN and the conserved cysteine causes the Ja helix to undock, become 85 disordered and move away from the core domain 38 39,40. Back in the dark, the covalent bond 86 with FMN decays, allowing the Ja helix to fold back into its tightly packed dark state 87 conformation39,40. 88 89 Several systems have exploited this large, light-triggered, conformational change of AsLOV2 to 90 confer light-dependence upon natural functions of different proteins. Insertion of AsLOV2 into 91 solvent-exposed loops of kinases, phosphatases and guanine exchange factors make their 92 enzymatic activities and downstream signaling events light-dependent41. A light-switchable Cas9 93 nuclease has also been developed using the AsLOV2 homolog from Rhodobacter sphaeroides, 94 RsLOV, in which the light-triggered conformational change of RsLOV dissociates a sterically 95 occluded RsLOV-Cas9 chimera homodimer to release Cas9 nuclease activity42. This approach 96 was also demonstrated in smaller proteins by fusing AsLOV2 to the small natural CRISPR 97 inhibitor AcrIIA4, resulting in a light-responsive anti-CRISPR system to control genome editing 98 with light43. The versatility of AsLOV2 to bestow light-dependent functionalities to chimeric 99 proteins in a variety of contexts, shown in these studies, encouraged us to use this light- 100 responsive protein to develop light controls in a protein binder. 101 102 Here we show that by fusing a monobody to the AsLOV2 domain we obtain a light-dependent 103 monobody, or OptoMonobody (OptoMB), whose binding affinity to the monobody’s cognate 104 protein target is controllable with light. Taking a structure-based protein engineering approach, 105 we inserted the AsLOV2 domain into structurally conserved, solvent exposed, loops of a 106 monobody that binds the SH2 domain of Abl kinase. We show that one of these chimeras 107 preferentially binds to the SH2 domain in the dark, both in vitro and in mammalian cells. We 108 harness the ~300-fold change in binding affinity between lit and dark states to implement light- 109 based protein purification44 using an OptoMB resin in what we call “light-controlled affinity 110 chromatography” (LCAC). This work, and an accompanying study reporting light-dependent 111 nanobodies or OptoNanobodies (OptoNBs)45, represent the first demonstrations of protein 112 binders with high affinity and selectivity, engineered with light-control of their binding activity.

4 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

113 The new class of light-responsive protein binders, or OptoBinders (OptoBNDRs), which in 114 principle can be engineered, screened, or selected to bind any protein of interest, have great 115 potential to be used in numerous new applications in biotechnology, synthetic biology, and basic 116 research. 117 118 Results 119 Design and selection of OptoMonobodies 120 121 To demonstrate the feasibility of developing a light-sensitive monobody, we chose the HA4

122 monobody, which binds with high affinity (Kd ~7nM) to the SH2 domain of the human Abl 123 kinase, in vitro and in vivo46. This is an interesting and valuable target, as many proteins 124 containing SH2 domains in general, and Abl kinase in particular, are involved in human health 125 and disease47–49. In addition, the availability of the crystal structure of HA4 bound to SH2 126 domain46 is a valuable resource for our rational protein engineering approach. Our strategy to 127 develop a light-sensitive HA4 was to design various chimeras of this monobody with the light- 128 oxygen-voltage- (LOV) sensing domain of Phototropin1 from Avena sativa, AsLOV2, and test 129 their ability to bind and release the SH2 domain depending on light conditions. 130 131 To build our chimeras, we used a truncated version of AsLOV2, which induces light-dependent 132 conformational changes in engineered nanobody chimeras more efficiently than its full-length 133 counterpart45. Our strategy was to insert the shortened AsLOV2 domain in all seven structurally- 134 conserved, solvent-exposed loops of HA4 (Fig. 1b). Given the large conformational change of 135 AsLOV2 triggered by light (Fig. 1a), our hypothesis was that the native conformation of the 136 monobody domain in some chimeras would be preserved in the dark, allowing it to bind to SH2, 137 but disrupted in the light, causing it to release its target. Guided by the crystal structure of HA4 138 bound to SH2 (PDB ID: 3k2m)46, we explored potential sites within the seven solvent-exposed 139 loops in HA4 where we could insert AsLOV2. We selected as many positions as possible in 140 each loop, avoiding those were we have reasons to believe the dark state conformation of 141 AsLOV2 would disrupt the core b-sheets of the monobody or interfere with the monobody-SH2 142 binding interaction. We also excluded positions where the light-triggered conformational change 143 of the AsLOV2 Ja helix might be impeded by clashes with the monobody core. After this

5 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

144 structural analysis, we selected 17 AsLOV2 insertion sites across all solvent exposed loops of 145 HA4, as well as N- or C-terminal fusions (Fig. 1b and Supplementary Table 1). 146 147

148 149 Fig. 1: Design and screens of AsLOV2-Monobody chimeras and final OptoMB. a, Light- 150 triggered conformational change of the Ja helix (green) of AsLOV2. b, Crystal structure of 151 Monobody HA4 (blue) bound to SH2 domain (gray). The monobody fold consists of two b- 152 sheets, bSh1(black) and bSh2 (light blue), and seven structurally conserved loops (red), where

6 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

153 AsLOV2 was inserted in our chimeras. Loop L4, where AsLOV2 is inserted in OptoMB, is 154 shown with an arrow. The cartoon below shows the relative size and location of loops (red), 155 including positions of AsLOV2 insertion (red lines) and insertion sites of identified light- 156 responsive chimeras: MS29 and SS58 (in OptoMB). c, Schematic diagram of pull-down screens 157 used to identify light-responsive chimeras. Co2+ agarose beads (pink) were used to immobilize

158 His6-YFP-SH2 (yellow-gray), which were then incubated with HA4-AsLOV2 chimeras (blue- 159 orange) in either dark or light. After washing and eluting with imidazole, the eluents were 160 resolved on SDS-PAGE, were differences in protein band intensity between samples exposed to 161 different light conditions reflect differences in chimera SH2 binding. d, Chimera pull-down 162 screen results showing SDS-PAGE protein bands of two chimeras that bind better in the dark 163 than in the light. These light-responsive chimeras have AsLOV2 inserted in either Loop L2 164 (MS29) or Loop L4 (SS58, SS59 and MS29). e, Energy-minimized structural model of the dark 165 conformation of OptoMB with AsLOV2 (orange with the Ja helix in green) inserted in position 166 SS58 of HA4 (blue and black), interacting with the SH2 domain (gray). 167 168 169 170 To find chimeras that can bind the SH2 domain in the dark but not in the light, we screened our 171 constructs using an in vitro pull-down assay (Fig. 1c). First, we produced an N-terminally His-

172 tagged fusion of yellow fluorescent protein (YFP) and SH2 domain (His6-YFP-SH2) in E. coli, 173 and immobilized it onto cobalt-charged agarose beads. We then incubated the beads with crude 174 extracts of E. coli expressing each of the different AsLOV2-HA4 chimeras, in either blue light or 175 darkness. After washing the beads under the same light conditions (see Methods), we eluted with 176 imidazole and resolved the products with denaturing polyacrylamide gel electrophoresis (SDS- 177 PAGE) to analyze the binding of each chimera in different light conditions (Fig. 1c,d). We 178 anticipated that chimeras that bind to SH2 preferentially in the dark would show a more intense 179 band on SDS-PAGE for samples that were incubated and washed in the dark, relative to the 180 samples treated in the light (Fig. 1c,d and Supplementary Fig. 1). 181 182 We found that AsLOV2 insertions in two different HA4 loops produce chimeras with the 183 expected behavior in our pull-down assays. One promising chimera has AsLOV2 inserted

7 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

184 between residues Met29 and Ser30 (a site we call MS29, following a naming system for sites 185 used in this study), located in loop L2 of HA4 (Fig. 1b,d and Supplementary Fig. 1). We only 186 saw an effect involving this loop when AsLOV2 was inserted at position MS29 and loop L2 was 187 shortened by removing the three surrounding amino acids (Ser30, Ser31 and Ser32, see 188 Supplementary Sequences). Another chimera with positive results has AsLOV2 inserted between 189 Ser58 and Ser59 (site SS58) located in loop L4 of HA4 (Fig. 1b,d). Insertions at other positions 190 within loop L4 (57YSSS60) show smaller degrees of variation in band intensity between beads 191 treated in the light versus in the dark (Fig. 1d and Supplementary Fig. 1). This suggests that loop 192 L4 is a “hot spot” for favorable orientations between AsLOV2 and the monobody to produce 193 light-responsive chimeras that switch between a conformation that allows target binding (in the 194 dark) and one that promotes target dissociation (in the light). Compared to AsLOV2 insertions at 195 other positions in loop L4, the insertion at SS58 shows the largest qualitative difference between 196 the chimera bound to beads in different light conditions (Fig. 1d and Supplementary Fig. 1). 197 Therefore, we chose this chimera to continue our study, naming it aSH2 OptoMonobody 198 (henceforth called OptoMB in this study). A structural model of this OptoMB (Fig. 1e and 199 Supplementary Movie 1) reveals that the orientation of AsLOV2 in this chimera (modeled in the 200 dark state) is compatible with SH2 binding and suggests a possible mechanism by which a light- 201 triggered conformational change of the Ja helix may disrupt the HA4 monobody to disfavor SH2 202 binding (see Discussion). 203 204 In vitro characterization of OptoMB 205 206 The in vitro light-dependent interaction of our OptoMB and SH2 domain (Fig. 2a) can be 207 visualized by fluorescence microscopy using immobilized OptoMB. We immobilized His-tagged 208 OptoMBs harboring a mutation in AsLOV2 (V416L) that extends the lifetime of the 50 209 photoactivated state (OptoMBV416L) onto Ni-charged agarose beads. Immobilized parental HA4 210 monobodies were used in parallel as a control. The monobody-coated beads were then incubated 211 with a fusion of yellow fluorescent protein and SH2 (YFP-SH2) in the dark until they reached 212 equilibrium and imaged over time using confocal microscopy in the presence or absence of blue 213 light (450 nm). For OptoMB-coated beads, light exposure induced a rapid decrease in YFP signal 214 on the surface of the bead within one imaging frame (2 min), consistent with light-triggered

8 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

215 dissociation of the SH2 domain (Fig. 2b upper panel and Supplementary Movie 2). In contrast, 216 no fluorescence change was observed under identical conditions for the control HA4-coated 217 beads (Fig. 2b lower panel and Supplementary Movie 2). Localized illumination could also be 218 used to restrict SH2 unbinding to a single bead in a crowded field (Supplementary Fig. 2a). In 219 this case, YFP fluorescence was rapidly and reversibly controlled for the illuminated bead but 220 not a nearby un-illuminated bead (Supplementary Fig. 2b and Supplementary Movie 3). These 221 results demonstrate that OptoMBs provide temporal, spatial, and reversible control over protein 222 binding. 223 224

225 226 Fig. 2: In vitro characterization of OptoMB. a, Schematic diagram of OptoMB and YFP-SH2 227 darkness-dependent interaction. In the dark OptoMB binds to SH2 domain of YFP-SH2 fusion;

9 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

228 upon blue light stimulation, the AsLOV2 domain disrupts the HA4 domain of OptoMB, reducing 229 its affinity to SH2. b, Time course of fluorescence microscopy images of YFP-SH2 binding to

230 agarose beads conjugated with OptoMBV416L (top panel) or HA4 as control (lower panel). 231 Starting with beads incubated in the dark, time course begins upon blue light stimulation (t = 0) , 232 followed by a sequence of images taken every 120 s for a total of 480 seconds (see also 233 Supplementary Movie X) c, Size exclusion chromatography profiles of OptoMB and YFP-SH2 234 interactions in light (blue line) and dark (black line). d, e, BLI measurements of binding (left) 235 and unbinding (right) activities of YFP-SH2 to immobilized OptoMB in the dark, d, (using

236 OptoMBC450V) or in the light, e, (using OptoMBV416I_G528A_N538E fused to SUMO tag). The 237 calculated rate and dissociation constants are shown below the BLI data. 238 239 240 241 The light dependency of OptoMB interactions with YFP-SH2 can also be analyzed in solution by 242 size exclusion chromatography (SEC). We prepared mixtures of purified YFP-SH2 and OptoMB 243 (or HA4 monobody, as control), in which OptoMB was added in excess (see Methods). We then 244 loaded each sample to a gel filtration column under continuous darkness or blue light 245 (Supplementary Fig. 2c). We found that for both the dark-incubated OptoMB sample (Fig. 2c) 246 and the HA4 monobody in either light condition (Supplementary Fig. 2d), the YFP-SH2 elutes 247 primarily as a monobody-bound complex. In contrast, the illuminated OptoMB sample shows a 248 higher retention time for the YFP-SH2-OptoMB complex and a larger proportion of the YFP- 249 SH2 eluting as a monomer on its own (Fig. 2c). Taken together, our bead-imaging and SEC data 250 are both consistent with a blue light-triggered reduction in the OptoMB-SH2 binding affinity, 251 both in solution and on protein-coated surfaces. 252 253 To quantify the changes in OptoMB-SH2 binding, we determined the kinetic rate constants and 254 binding affinity in different light conditions. In these assays we took advantage of three classes 255 of mutations in AsLOV2 to vary properties of the OptoMB binding switch. Bio-layer 256 interferometry uses visible light for measuring changes in binding, so we first prepared a light- 257 insensitive OptoMB by introducing the well-characterized C450V mutation in AsLOV2 that 36,51 258 renders it light-insensitive (OptoMBC450V). Conversely, we ensured that illumination could

10 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

259 drive efficient conversion to the lit state by generating additional OptoMB variants with 260 mutations that extend the lifetime of the lit state after illumination from ~80 to 821 or up to 261 4300 sec in vitro (AsLOV2 V416I49 or V416L52 respectively). Finally, we added the AsLOV2

262 G528A and N538E mutations (to make the triple mutant OptoMBV416I_G528A_N538E), which have 263 been reported to stabilize the dark state conformation and potentially decrease leakiness in the 264 absence of illumination53,54. We fit the resulting binding and dissociation data to a mass-action 265 kinetic binding model (Fig. 2d,e and Supplementary Fig. 2e,f), allowing us to obtain estimates of

266 the rate constants of binding (kon) and unbinding (koff) as well as the overall dissociation constant

267 (Kd) of the OptoMB-SH2 interaction in different light conditions (Supplementary Table 2). 268 269 We found that the binding affinity of OptoMB to SH2 changes dramatically when switching 270 from dark to light conditions. The dissociation constant of the OptoMB-SH2 interaction in the

271 dark is Kd = 0.07 µM, comparable to that of the HA4 monobody. However, in the light it

272 increases to Kd = 11.2 µM or Kd =21.7 µM, depending on which lit-stabilized OptoMB variant 273 we use. This amounts to a 160-310 fold-change in binding affinity upon switching light 274 conditions, which explains the light-dependent behaviors observed in the bead-imaging and SEC

275 experiments above. These changes in Kd arise primarily from the decrease in the OptoMB-SH2

276 binding rate constant (kon) in the light, although we also observe an increase in the unbinding rate

277 constant (koff) (Supplementary Table 2). These data are consistent with a light-induced change in 278 the conformation of the OptoMB that disrupts the binding interface, substantially decreasing the 279 likelihood of a productive association between the lit-state OptoMB and its target SH2.

280 Moreover, the dominant role of kon variation in light-induced changes in binding affinity is 281 consistent with our proposed OptoMB functional model, which opens the door to further 282 engineering and optimization (see Discussion). 283 284 Light-controlled affinity chromatography (LCAC) with immobilized OptoMB 285 286 We reasoned that the substantial change in OptoMB binding affinity could open the door to 287 purifying a protein of interest simply by shifting illumination conditions (Fig. 3a), a procedure 288 that we termed “light-controlled affinity chromatography” (LCAC). We immobilized two 289 variants of His-tagged OptoMBs harboring either the wild-type AsLOV2 or the triple-mutant

11 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

290 (OptoMBV416I_G528A_N538E) described above, onto Co-charged Agarose beads to make an αSH2- 291 OptoMB affinity resin. We then incubated OptoMB-coated beads with crude lysate from E. coli 292 overexpressing YFP-SH2. After washing thoroughly in the dark (see Methods), we eluted with 293 blue light either in batch (Fig. 3b) or in a column (Supplementary Fig. 3). After elution, beads 294 were washed with imidazole to recover any remaining protein bound to the beads in order to 295 estimate the capacity and yields of the resin. With these initial LCAC purification trials, we 296 achieved 95-98% purity in a single step, with yields ranging from 18-30% and binding capacities 297 from 112-145 nmol (4.5-6 mg) of SH2-tagged YFP per mL of OptoMB resin, depending on the 298 OptoMB variant used (Supplementary Table 3). 299 300 To test whether LCAC could be applied to larger and more complex proteins, we also used it to 301 purify the main pyruvate decarboxylase from Saccharomyces cerevisiae, Pdc1p. This enzyme 302 catalyzes the decarboxylation of pyruvate to acetaldehyde for ethanol fermentation, and is 303 composed of a homotetramer of 61 kDa monomers55, significantly larger than YFP. We fused the 304 SH2 domain to the N-terminus of Pdc1p (SH2-PDC1) and performed LCAC to purify it from

305 crude E. coli lysate, as described above, using a resin coated with OptoMBV416I_G528A_N538E. This 306 procedure enabled purification of Pdc1p to 96% purity with a 39% yield (Fig. 3c and 307 Supplementary Table 3). It is noteworthy that this purification works considerably well, despite 308 the potential binding avidity of Pdc1p tetramers that would be predicted to increase the protein’s 309 apparent affinity in both the light and dark. These results demonstrate that LCAC can be applied 310 to purify relatively large proteins with quaternary structures of up to at least 300 kDa (including 311 the fused SH2 domain), achieving a high degree of purity and an acceptable yield. Our results 312 with YFP-SH2 and SH2-PDC1 further demonstrate that OptoMB-assisted purification is 313 compatible with both N- and C-terminal SH2 tags. 314 315 316 317 318 319 320

12 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

321

322 323 Fig. 3: Light-Controlled Affinity Chromatography (LCAC) to purify SH2-tagged proteins 324 using cobalt-immobilized OptoMB. a, Schematic diagram of functionalization of LCAC

325 agarose beads using His6-OptoMBV416I_G528A_N538E fused to an N-terminus SUMO tag (S) (bound 326 to Co2+ beads) and LCAC procedure involving incubation of crude E. coli extract, and washing 327 in the dark, followed by elution with blue light in batch or column (Supplementary Fig. 3). 328 Finally, SDS-PAGE was used to resolve the fractions from each purification step. b, c, SDS 329 PAGE gel of LCAC-purified YFP-SH2 (b), and SH2-PDC1 (c). Molecular weight markers (M), 330 lysate (L), unbound flow through (FT), washing step in the dark (Dark Wash), two consecutives 331 light elution aliquots (Light Elution 1 and 2), and the imidazole elution (I) were resolved in 12% 332 SDS-PAGE. 333 334 335 336 While metal-affinity beads are effective at immobilizing OptoMB for LCAC (Fig. 3 and 337 Supplementary Fig. 3), they may be incompatible with some protein purification methods. Thus, 338 to determine whether an alternative resin could be used for LCAC, we immobilized our

13 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

339 OptoMBV416I_G528A_N538E onto cyanogen bromide-activated sepharose beads (CNBr-beads), which 340 immobilizes proteins by making covalent bonds with its primary amines (see Methods). 341 Following the same purification protocol as above, we found that CNBr-beads are also effective 342 at purifying both YFP and Pdc1p (Supplementary Fig. 4). A single step of CNBr-based 343 purification achieved yields above 40% and purity of 96.7–99.8%, surpassing any other LCAC 344 method tested (Supplementary Table 3). These gains are likely related to the reduced non- 345 specific binding of E. coli proteins to CNBr-sepharose and covalent OptoMB attachment which 346 allowed for more extensive washing at higher salt concentrations. We observed that the total 347 loading capacity is not as high as that of OptoMB-coated Co-agarose (Supplementary Table 3), 348 probably because random crosslinking to the CNBr-beads inactivates a significant fraction of the 349 OptoMB by occluding its binding surface to SH2. Although much could still be done to further 350 improve LCAC by optimizing the resin, method, or amino acid sequence of the OptoMB, these 351 experiments demonstrate the feasibility of a practical in vitro application of light-responsive 352 monobodies for protein purification. 353 354 Light-dependent OptoMB binding in vivo 355 356 We have shown that OptoMB-SH2 binding can be controlled with light in vitro; as a final test of 357 this system, we assessed whether similar control can be achieved in live mammalian cells where 358 monobodies have been reported to work33. We transduced HEK293T cells with lentiviral vectors 359 encoding a membrane-localized, fluorescent SH2 target protein (SH2-mCherry-CAAX) and 360 cytosolic fluorescent OptoMB (OptoMB-irFP), reasoning that a light-dependent change in SH2- 361 OptoMB binding would cause the OptoMB to redistribute between the cytosol and plasma 362 membrane (PM) (Fig. 4a), as has been observed for conventional optogenetic protein-protein 363 interactions in previous studies56–58. As a control, we expressed irFP-labeled HA4 monobody 364 instead of OptoMB, which would be expected to bind to the membrane-localized SH2-mCherry- 365 CAAX regardless of illumination conditions. 366 367 Fluorescence imaging confirmed the PM localization of SH2-mCherry-CAAX (Fig. 4b,c left 368 panels) as well as constitutively PM-bound HA4-irFP (Fig. 4b). In contrast, we observe a light- 369 dependent shift in OptoMB localization (Fig. 4c), with PM enrichment in the dark and rapid

14 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

370 redistribution to the cytosol upon light stimulation. Applying cycles of light and darkness further 371 revealed that light-controlled binding is fully reversible intracellularly (Fig. 4d; Supplementary 372 Movie 4). These experiments demonstrate functional photoswitching of OptoMB binding inside 373 cells, opening the door to its in vivo application. 374

375 376 Fig. 4: In vivo characterization of OptoMB using mammalian cells. a, Schematic diagram of 377 the membrane-binding assay used, where irFP-tagged OptoMB (blue-orange-purple), binds to a 378 fusion of SH2 (gray), mCherry (red), and CAAX (black linker), anchored to the plasma 379 membrane (PM) of HEK 293T cells. In the dark, OptoMB-irFP binds to the membrane-bound 380 SH2 enhancing irFP signal at the PM, reducing its cytosolic signal. In blue light, OptoMB

15 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

381 releases SH2, causing a reduction in irFP signal at the membrane and an increase in the cytosol. 382 b, HEK 293T cells expressing SH2-mCherry-CAAX and irFP-tagged HA4 monobody, as 383 control, imaged in the dark or blue light. Left panel shows mCherry fluorescence of SH2 fusion 384 anchored to the PM; central and right panels show HA4-irFP fluorescence in the dark and after 385 30 s of blue light stimulation, respectively, showing irFP-HA4 localized to the cytosol in either 386 light condition. c, HEK 293T cells expressing SH2-mCherry-CAAX and OptoMB-iRFP imaged 387 in the dark or blue light. Left panel, shows mCherry fluorescence of the SH2 fusion anchored to 388 the PM. Central and right panels show HA4-irFP fluorescence in the dark and after 30 s of blue 389 light stimulation, respectively, showing OptoMB enriched at the PM in the dark and in the 390 cytosol in the light. d, Cytosolic irFP-HA4 fluorescence changing over time due to periodic 391 pulses of blue light (blue sections). The fluorescence is expressed in percentage from the original 392 value of cells exposed to the dark. Curve and shaded regions indicate mean ± SD for at least 10 393 cells respectively. White scale bar in the cell images is 10 µm. 394 395 396 397 Discussion 398 Here we show that by taking a rational protein engineering approach it is possible to develop a 399 light-switchable monobody (OptoMB). Our OptoMB design strategy was based on the fusion of 400 a light-switchable AsLOV2 domain to a structurally conserved loop of the H4A monobody 401 which binds with high affinity and selectivity to the SH2 domain of the human Abl kinase. We 402 find that this chimera binds to SH2 in the dark with similar affinity as its parental monobody 403 H4A; but in blue light its binding affinity drops 150- to 300-fold relative to the dark state 404 (Supplementary Table 2). Furthermore, the light responsiveness of OptoMB is effective to 405 control binding to proteins fused to SH2 at either their N- or C- terminus. OptoMBs, along with 406 the light-dependent nanobodies (OptoNBs) we describe in an accompanying study45, belong to a 407 new class of light-dependent protein binders we call OptoBinders (OptoBNDRs), which offer 408 promising new in vivo and in vitro applications. 409 410 A close inspection of the structural model of OptoMB and binding measurements suggest a 411 possible mechanism for the light-dependent binding affinity of OptoMB. The monobody protein

16 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

412 fold consists of two antiparallel b-sheets (bSh1 and bSh2) that interact with each other to form 413 the protein core (Fig. 1b). In our original chimera screens, we inserted AsLOV2 in all 414 intervening loops within (L1, L3, L6 and L7) and between (L2, L4 and L5) these b-sheets. The 415 only chimeras that show a change in binding affinity in different light conditions are those with 416 AsLOV2 inserted in loops L2 and L4, with L4 being the only loop with positive results at 417 multiple insertion sites. Both of these loops connect bSh1 and bSh2 with each other, suggesting 418 that chimeras involving either loop may act by a similar mechanism, where the light-triggered 419 conformational change of the Ja helix pulls the bSh1 and bSh2 apart from each other. Chimeras 420 with AsLOV2 inserted in loop L5, located at the opposite side of the bSh1-bSh2 interaction 421 relative to L2 and L4, show equally faint SDS-PAGE bands in either light condition, suggesting 422 limited expression in E. coli, or weak binding to SH2 independently of light. The consequence of 423 pulling on bSh1 and bSh2 from loops L2 or L4 is probably at least a partial disruption of the 424 interactions and angle between them. This in turn would likely change the curvature of bSh2, 425 which defines the shape of the paratope-like surface of H4A and its specific binding interactions 426 with SH246. Interestingly, because the conserved fold of monobodies always includes bSh1 and 427 bSh2 interactions31, this light-triggered disruption of the binding surface may be transferable to 428 other monobodies. It also suggests that mutating residues involved in bSh1-bSh2 interactions 429 may provide some opportunities to tune the photoswitchable behavior of the OptoMB, by 430 stabilizing either the dark- or lit-state conformations. 431 432 This model of light-induced disruption of the monobody’s target-binding site is also consistent 433 with our measurements of binding kinetics. We originally reasoned that light stimulation might 434 strain the OptoMB-SH2 interaction causing them to dissociate without necessarily inducing 435 dramatic changes in binding site accessibility, thus predicting a light-induced change in the

436 OptoMB-SH2 off-rate (koff). While we do observe as much as a 10-fold increase in koff in the 437 light, consistent with this hypothesis, we found that most of the change in affinity comes from as

438 much as a 46-fold change in the binding rate (kon) between dark and light conditions 439 (Supplementary Table 2). This suggests that the binding site is disrupted enough to inhibit 440 OptoMB-SH2 association in the first place. Yet despite what is likely to be a substantial 441 conformational change in the OptoMB, we find that productive binding is fully reversible in

17 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

442 vitro (Supplementary Fig. 2b and Supplementary Movie 3) and in vivo (Fig. 4d and 443 Supplementary Movie 4). Overall, these data are consistent with a large change in the overall 444 orientation or conformation of the bSh1-bSh2 sheets, disrupting binding but without driving 445 irreversible protein misfolding. It is possible that the small size of the monobody domain, the 446 short range of interactions between bSh1 and bSh2, and the lack of disulfide bonds in the 447 monobody fold facilitate this high efficiency of interconversion between binding states. 448 449 Comparison of our binding data to a maximum theoretical value that assumes perfect 450 transmission of energy between AsLOV2 and the monobody’s binding interface suggests that 451 light-induced changes in OptoMB could be further improved. Previous studies have measured 452 the free energy available from the dark-to-light conformational change of AsLOV2 using NMR 453 spectroscopy59 and developed analytical models to study equilibrium constants of the dark- and 454 lit-states of AsLOV253. These studies predict that 3.8 kcal/mol of energy is transmitted from the 455 absorption of one photon to structural rearrangements in AsLOV2. If all of this energy was 456 transmitted to a change in OptoMB binding, it would result in a ~600-fold difference in binding

457 affinity between the dark and lit states. Our BLI experiments measured a change in Kd values of 458 150- to 300-fold, which is within the same order of magnitude of this maximum theoretical 459 prediction, indicating that the light-induced conformational change of AsLOV2 is efficiently 460 transmitted to disrupt the SH2 binding surface of the monobody domain. Nevertheless, it also 461 suggests that further improvements might be achieved by optimizing the insertion site or linkers 462 between the AsLOV2 and monobody, or by engineering the H4A domain to improve the light- 463 induced allosteric coupling between the AsLOV2 domain and SH2-binding site. It is also 464 possible that performance for particular applications might be improved by tuning the OptoMB- 465 SH2 affinity to either weaker or tighter values. Finally, we note that the AsLOV2 domain might 466 also be altered to tune the efficiency of the light-induced conformational change, for instance by 467 further stabilizing the lit or dark state conformations or by altering its photoswitching kinetics 468 (Fig. 3b, c and Supplementary Table 2). In principle, these changes could increase the overall 469 energy beyond the 3.8 kcal/mol measured for the wild-type AsLOV2 domain. The structural 470 targets set for optimization and the expected results will of course vary depending on the 471 particular application objectives and the individual OptoMB/ligand pair in question. 472

18 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

473 The performance of our light-dependent OptoMB/SH2 binding pair is comparable to that of 474 other optical dimerization systems previously used to control transcription, protein-protein 475 interactions, or protein localization60–62. However, these previous systems have been developed 476 by fusing proteins of interest to specific light-dependent interaction partners (such as PhyB/PIF3 477 or Cry2/CIB) evolved in photosensitive organisms. While the demonstrations we used in this 478 study rely on fusing SH2 to different proteins, the true potential of this technology is not so much 479 to replace the light-responsive tags used in previous optogenetic systems with SH2 and OptoMB, 480 but in the possibility of engineering other monobodies against different targets of interest to 481 make their specific interaction light-dependent. The relative simplicity of the monobody fold and 482 the likely structural mechanism that confers light-dependent binding makes it reasonable to 483 expect that other monobodies could be engineered to be light-dependent; starting from inserting 484 AsLOV2 in loop L4 and then optimizing the precise position of the insertion site and the linkers. 485 Monobodies could in principle be designed and selected to bind any protein of interest6,31 or to 486 different epitopes on a single a target31. Our approach may thus be useful to design reversible 487 interactions to a protein of interest without the need for a binding tag, which for some proteins 488 may be impractical, not possible, or interfere with their natural activity. 489 490 References 491 492 1. Skerra, A. Alternative binding proteins: Anticalins - Harnessing the structural plasticity of 493 the lipocalin ligand pocket to engineer novel binding activities. FEBS J. 275, 2677–2683 494 (2008). 495 2. Škrlec, K., Štrukelj, B. & Berlec, A. Non-immunoglobulin scaffolds: A focus on their 496 targets. Trends in Biotechnology 33, 408–418 (2015). 497 3. Yu, X., Yang, Y.-P., Dikici, E., Deo, S. K. & Daunert, S. Beyond Antibodies as Binding 498 Partners: The Role of Antibody Mimetics in Bioanalysis. Annu. Rev. Anal. Chem. 10, 499 293–320 (2017). 500 4. Frejd, F. Y. & Kim, K. T. Affibody molecules as engineered protein drugs. Exp. Mol. 501 Med. 49, e306-8 (2017). 502 5. Plückthun, A. Designed Ankyrin Repeat Proteins (DARPins): Binding Proteins for 503 Research, Diagnostics, and Therapy. Annu. Rev. Pharmacol. Toxicol. 55, 489–511 (2015).

19 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

504 6. Sha, F., Salzman, G., Gupta, A. & Koide, S. Monobodies and other synthetic binding 505 proteins for expanding protein science. Protein Science 26, 910–924 (2017). 506 7. Ingram, J. R., Schmidt, F. I. & Ploegh, H. L. Exploiting Nanobodies’ Singular Traits. 507 Annu. Rev. Immunol. 36, (2018). 508 8. Sormanni, P., Aprile, F. A. & Vendruscolo, M. Third generation antibody discovery 509 methods:: In silico rational design. Chemical Society Reviews 47, 9137–9157 (2018). 510 9. Helma, J., Cardoso, M. C., Muyldermans, S. & Leonhardt, H. Nanobodies and 511 recombinant binders in cell biology. J. Cell Biol. 209, 633–644 (2015). 512 10. Sullivan, M. A., Wentworth, T., Kobie, J. J. & Sanz, I. Anti-idiotypic monobodies for 513 immune response profiling. Methods 58, 62–68 (2012). 514 11. Hu, Y., Liu, C. & Muyldermans, S. Nanobody-based delivery systems for diagnosis and 515 targeted tumor therapy. Front. Immunol. 8, (2017). 516 12. Bannas, P., Hambach, J. & Koch-Nolte, F. Nanobodies and nanobody-based human heavy 517 chain antibodies as antitumor therapeutics. Front. Immunol. 8, 1–13 (2017). 518 13. Vazquez-Lombardi, R. et al. Challenges and opportunities for non-antibody scaffold 519 drugs. Drug Discov. Today 20, 1271–1283 (2015). 520 14. Bielser, J. M., Wolf, M., Souquet, J., Broly, H. & Morbidelli, M. Perfusion mammalian 521 cell culture for recombinant protein manufacturing – A critical review. Biotechnology 522 Advances 36, 1328–1340 (2018). 523 15. Tripathi, N. K. Production and Purification of Recombinant Proteins from Escherichia 524 coli . ChemBioEng Rev. 3, 116–133 (2016). 525 16. Gronemeyer, P., Ditz, R. & Strube, J. Trends in Upstream and Downstream Process 526 Development for Antibody Manufacturing. Bioengineering 1, 188–212 (2014). 527 17. Richards, D. A. Exploring alternative antibody scaffolds: Antibody fragments and 528 antibody mimics for targeted drug delivery. Drug Discov. Today Technol. 30, 35–46 529 (2018). 530 18. Nishikido, T. & Ray, K. K. Non-antibody Approaches to Proprotein Convertase Subtilisin 531 Kexin 9 Inhibition: siRNA, Antisense Oligonucleotides, Adnectins, Vaccination, and New 532 Attempts at Small-Molecule Inhibitors Based on New Discoveries. Front. Cardiovasc. 533 Med. 5, (2019). 534 19. Activity, A. et al. crossm Discovery and Characterization of a Novel CD4-Binding

20 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

535 Adnectin with Potent. 1–19 (2017). 536 20. Rodrigo, G., Gruvegård, M. & Van Alstine, J. Antibody Fragments and Their Purification 537 by Protein L Affinity Chromatography. Antibodies 4, 259–277 (2015). 538 21. Jin, J. et al. Accelerating the clinical development of protein-based vaccines for malaria 539 by efficient purification using a four amino acid C-terminal ‘C-tag’. Int. J. Parasitol. 47, 540 435–446 (2017). 541 22. Pabst, T. M. et al. Camelid VHH affinity ligands enable separation of closely related 542 biopharmaceuticals. Biotechnol. J. 12, 1–10 (2017). 543 23. Götzke, H. et al. A rationally designed and highly versatile epitope tag for nanobody- 544 based purification, detection and manipulation of proteins. bioRxiv 640771 (2019). 545 doi:10.1101/640771 546 24. Zhao, W., Pferdehirt, L. & Segatori, L. Quantitatively Predictable Control of Cellular 547 Protein Levels through Proteasomal Degradation. ACS Synth. Biol. 7, 540–552 (2018). 548 25. Ponnambalam, S. et al. Affimer proteins are versatile and renewable affinity reagents. 549 Elife 6, 1–35 (2017). 550 26. Zorba, A. et al. Allosteric modulation of a human protein kinase with monobodies. Proc. 551 Natl. Acad. Sci. U. S. A. 116, 13937–13942 (2019). 552 27. Spencer-Smith, R. et al. Inhibition of RAS function through targeting an allosteric 553 regulatory site. Nat. Chem. Biol. 13, 62–68 (2017). 554 28. Hamers-Casterman, C. et al. Naturally occurring antibodies devoid of light chains. Nature 555 363, 446–448 (1993). 556 29. Holliger, P. & Hudson, P. J. Engineered antibody fragments and the rise of single 557 domains. Nat. Biotechnol. 23, 1126–1136 (2005). 558 30. Koide, A., Bailey, C. W., Huang, X. & Koide, S. The fibronectin type III domain as a 559 scaffold for novel binding proteins. J. Mol. Biol. 284, 1141–1151 (1998). 560 31. Koide, A., Wojcik, J., Gilbreth, R. N., Hoey, R. J. & Koide, S. Teaching an old scaffold 561 new tricks: Monobodies constructed using alternative surfaces of the FN3 scaffold. J. Mol. 562 Biol. 415, 393–405 (2012). 563 32. Tolcher, A. W. et al. Phase I and pharmacokinetic study of CT-322 (BMS-844203), a 564 targeted adnectin inhibitor of VEGFR-2 based on a domain of human fibronectin. Clin. 565 Cancer Res. 17, 363–371 (2011).

21 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

566 33. Gupta, A. et al. Facile target validation in an animal model with intracellularly expressed 567 monobodies. Nat. Chem. Biol. 14, 895–900 (2018). 568 34. Schmit, N. E., Neopane, K. & Hantschel, O. Targeted Protein Degradation through 569 Cytosolic Delivery of Monobody Binders Using Bacterial Toxins. ACS Chem. Biol. 14, 570 916–924 (2019). 571 35. Harper, S. M., Neil, L. C. & Gardner, K. H. Structural Basis of a Phototropin i l l l l l l. 572 301, 1541–1545 (2003). 573 36. Zayner, J. P., Antoniou, C. & Sosnick, T. R. The amino-terminal helix modulates light- 574 activated conformational changes in AsLOV2. J. Mol. Biol. 419, 61–74 (2012). 575 37. Peter, E., Dick, B. & Baeurle, S. A. Mechanism of signal transduction of the LOV2-Jα 576 photosensor from Avena sativa. Nat. Commun. 1, 122–127 (2010). 577 38. Crosson, S. & Moffat, K. Structure of a flavin-binding plant photoreceptor domain: 578 Insights into light-mediated signal transduction. Proc. Natl. Acad. Sci. U. S. A. 98, 2995– 579 3000 (2001). 580 39. Konold, P. E. et al. Unfolding of the C-Terminal Jα Helix in the LOV2 Photoreceptor 581 Domain Observed by Time-Resolved Vibrational Spectroscopy. J. Phys. Chem. Lett. 7, 582 3472–3476 (2016). 583 40. Harper, S. M., Christie, J. M. & Gardner, K. H. Disruption of the LOV-Jα helix interaction 584 activates phototropin kinase activity. Biochemistry 43, 16184–16192 (2004). 585 41. Dagliyan, O. et al. Engineering extrinsic disorder to control protein activity in living cells. 586 Science (80-. ). 354, 1441–1444 (2016). 587 42. Richter, F. et al. Engineering of temperature- and light-switchable Cas9 variants. Nucleic 588 Acids Res. 44, 10003–10014 (2016). 589 43. Bubeck, F. et al. Engineered anti-CRISPR proteins for optogenetic control of CRISPR– 590 Cas9. Nat. Methods 15, 924–927 (2018). 591 44. Hörner, M. et al. Light-controlled affinity purification of protein complexes exemplified 592 by the resting ZAP70 interactome. Front. Immunol. 10, 1–12 (2019). 593 45. Gil, A. A. et al. Optogenetic control of protein binding using light-switchable nanobodies. 594 9881, (2019). 595 46. Wojcik, J. et al. A potent and highly specific FN3 monobody inhibitor of the Abl SH2 596 domain. Nat. Struct. Mol. Biol. 17, 519–527 (2010).

22 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

597 47. Filippakopoulos, P., Müller, S. & Knapp, S. SH2 domains: modulators of nonreceptor 598 tyrosine kinase activity. Current Opinion in Structural Biology 19, 643–649 (2009). 599 48. Marengere, L. E. M. & Pawson, T. Structure and function of SH2 domains. J. Cell Sci. 600 1994, 97–104 (2013). 601 49. Morlacchi, P., Robertson, F. M., Klostergaard, J. & McMurray, J. S. Targeting SH2 602 domains in breast cancer. Future Med. Chem. 6, 1909–1926 (2014). 603 50. Kawano, F., Aono, Y., Suzuki, H. & Sato, M. Fluorescence imaging-based high- 604 throughput screening of fast- and slow-cycling LOV proteins. PLoS One 8, (2013). 605 51. Gil, A. A. et al. Femtosecond to Millisecond Dynamics of Light Induced Allostery in the 606 Avena sativa LOV Domain. J. Phys. Chem. B 121, 1010–1019 (2017). 607 52. Zoltowski, B. D., Vaccaro, B. & Crane, B. R. Mechanism-based tuning of a LOV domain 608 photoreceptor. Nat. Chem. Biol. 5, 827–834 (2009). 609 53. Strickland, D. et al. Rationally improving LOV domain-based photoswitches. Nat. 610 Methods 7, 623–626 (2010). 611 54. Song, S. H. et al. Modulating LOV domain photodynamics with a residue alteration 612 outside the chromophore binding site. Biochemistry 50, 2411–2423 (2011). 613 55. Lu, G., Dobritzsch, D., Baumann, S., Schneider, G. & König, S. The structural basis of 614 substrate activation in yeast pyruvate decarboxylase. A crystallographic and kinetic study. 615 Eur. J. Biochem. 267, 861–868 (2000). 616 56. Levskaya, A., Weiner, O. D., Lim, W. A. & Voigt, C. A. Spatiotemporal control of cell 617 signalling using a light-switchable protein interaction. Nature 461, 997–1001 (2009). 618 57. Kennedy, M. J. et al. Rapid blue-light-mediated induction of protein interactions in living 619 cells. Nat. Methods 7, 973–975 (2010). 620 58. Yazawa, M., Sadaghiani, A. M., Hsueh, B. & Dolmetsch, R. E. Induction of protein- 621 protein interactions in live cells using light. Nat. Biotechnol. 27, 941–945 (2009). 622 59. Yao, X., Rosen, M. K. & Gardner, K. H. Estimation of the available free energy in a 623 LOV2-Jα photoswitch. Nat. Chem. Biol. 4, 491–497 (2008). 624 60. Salinas, F., Rojas, V., Delgado, V., Agosin, E. & Larrondo, L. F. Optogenetic switches for 625 light-controlled gene expression in yeast. Appl. Microbiol. Biotechnol. (2017). 626 doi:10.1007/s00253-017-8178-8

23 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

627 61. Zhao, E. M. et al. Optogenetic regulation of engineered cellular metabolism for microbial 628 chemical production. Nature 555, 683–687 (2018). 629 62. Yumerefendi, H. et al. Control of Protein Activity and Cell Fate Specification via Light- 630 Mediated Nuclear Translocation. PLoS One 10, e0128443 (2015). 631 63. Goulas, T. et al. The pCri system: A vector collection for recombinant protein expression 632 and purification. PLoS One 9, (2014). 633 64. Halavaty, A. S. & Moffat, K. N- and C-terminal flanking regions modulate light-induced 634 signal transduction in the LOV2 domain of the blue light sensor phototropin 1 from Avena 635 sativa. Biochemistry 46, 14001–14009 (2007). 636 65. Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of Coot. 637 Acta Crystallogr. Sect. D Biol. Crystallogr. 66, 486–501 (2010). 638 66. Krieger, E. et al. Improving physical realism, stereochemistry, and side-chain accuracy in 639 homology modeling: Four approaches that performed well in CASP8. Proteins Struct. 640 Funct. Bioinforma. 77, 114–122 (2009). 641 67. Studier, F. W. Protein production by auto-induction in high density shaking cultures. 642 Protein Expr. Purif. 41, 207–34 (2005). 643 68. Schindelin, J. et al. Fiji: An open-source platform for biological-image analysis. Nat. 644 Methods 9, 676–682 (2012). 645 69. Alexandre, M. T. A., Arents, J. C., Van Grondelle, R., Hellingwerf, K. J. & Kennis, J. T. 646 M. A base-catalyzed mechanism for dark state recovery in the Avena sativa phototropin-1 647 LOV2 domain. Biochemistry 46, 3129–3137 (2007). 648 649 Methods 650 Plasmid construction of chimeras for bacterial expression 651 One-step isothermal assembly reactions (Gibson assembly) were performed using previously 652 described methods61. The monobody HA4 and the SH2 domain (codon optimized for E. coli 653 expression) were ordered as gBlocks from Integrated DNA Technologies (IDT) containing 654 homology arms. The following vectors from the pCri system63 were used: pCri-7b for constructs 655 without a 6x-histidine tag; pCri-8b for constructs with a 6x-histidine tag (N-terminus), and pCri- 656 11b for constructs with both SUMO and 6x-histidine tags (N-terminus). As the pCri vectors 657 contain YFP, the synthetized SH2 domain was inserted into pCri-7b and 8b (previously

24 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

658 linearized with XhoI) by Gibson assembly to construct EZ-L664 and EZ-L703 (see 659 Supplementary table 4). Monobody HA4 was inserted into pCri-7b; the vector was digested 660 (opened) with NheI and XhoI and Gibson-assembled to build the template (EZ-L663) used for 661 the AsLOV2 insertions. A stop codon was added before the in-frame (C-terminus) 6x-histidine 662 tag of pCri-7b. The AsLOV2 domain (residues 408-543) was either amplified by PCR from 663 previous constructs45 (wt AsLOV2), or synthesized as IDT gBlocks (AsLOV2 mutants). To 664 insert AsLOV2 into the monobody, the backbone from the initial construct containing HA4 (EZ- 665 L663) was PCR amplified, using Takara Hifi PCR premix, starting from the insertion positions 666 (Supplementary Table 1) that were selected (and adding homology arms to AsLOV2). Next, 667 chimeras were finally assembled mixing each of the amplified products of the backbone PCR 668 from EZ-L663 with the AsLOV2 domain obtained from either PCR amplification (wt AsLOV2) 669 or synthesized by gBlocks (AsLOV2 mutants). SUMO tags were added by inserting (with 670 Gibson assembly) the PCR product of the full-length chimeras (with homology arms) into the 671 pCri-11b plasmid, previously opened with NheI and XhoI. PDC1 was amplified from S. 672 cerevisiae (S288C) genomic DNA using PCR, also with homology arms, and the construct (EZ- 673 L886) was built via Gibson assembly (3 fragments) with digested pCri-7b (NheI and XhoI) and 674 the PCR product of SH2 amplified from EZ-L664 (with homology arms). All constructs 675 (Supplementary Table 4) were sequenced by Genewiz and all protein sequences are available in 676 Supplementary Sequences. We used chemically competent DH5α to clone all vectors. After 677 verifying the plasmid sequence, vectors were used to transform chemically competent BL21 678 (DE3) or Rosetta strains for protein expression. 679 680 Construction of the structural model of OptoMB 681 To build the structural model of the HA4-AsLOV2 chimera (OptoMB) interacting with the SH2 682 domain (Fig. 1e), a shortened version45 (residues 408-543) of the AsLOV2 domain (PDB ID: 683 2V1A)64 was manually inserted to residues S58 and S59 of the monobody HA4, using the 684 program Coot65. The crystal structure of HA4 in complex with the SH2 domain (PDB ID: 685 3k2m)46 was used as template. After a manual adjustment, an energy minimization of the HA4- 686 AsLOV2 chimera was carried out with the website version of YASARA66 687 (http://www.yasara.org/minimizationserver.htm). 688

25 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

689 Plasmid Construction for mammalian cells 690 Constructs for mammalian cell experiments were cloned using backbone PCR and inFusion 691 (Clontech). Monobody HA4 or OptoMB variants were PCR amplified and Gibson-assembled 692 from bacterial plasmids (described above) into a pHR vector with a C-terminal irFP fusion 693 (Addgene #111510). The SH2 domain was amplified from EZ-L664 using PCR, and Gibson- 694 assembled into a pHR vector containing a C-terminal mCherry-CAAX fusion tag (Addgene 695 #50839). Stellar E. coli cells (TaKaRa) were transformed with these plasmids for amplification 696 and DNA storage. All plasmids were sequenced by Genewiz to verify quality. 697 698 Lentivirus Production and Transduction 699 HEK 293T cells were plated on a 12-well plate, reaching 40% confluency the next day. The cells 700 were then co-transfected with the corresponding pHR plasmid and lentiviral packaging plasmids 701 (pMD and CMV) using Fugene HD (Promega). Cells were incubated for ~ 48 hours and virus 702 was collected and filtered through a 0.45 mm filter. In addition, 2 µL of polybrene and 40 µL of 703 HEPES were added to the 2 mL viral solution. For infection, HEK 293T cells were plated on a 6- 704 well plate, allowed to adhere and reach 40% confluency. At that time, the cells were infected 705 with 200-500 µL of viral solution. All imaging was done at least 48 hours post infection. Cells 706 were cultured in DMEM media with 10% FBS, penicillin (100 U/mL), and streptomycin (0.1 707 mg/mL). 708 709 Screening for light-responsive monobodies

710 A 6x-histidine tagged fusion of YFP and SH2 (His6-YFP-SH2) was grown in 500 mL of 711 autoinduction67 media + Kanamycin (Kan)(50 µg/mL) for 16 hours at 30°C. Monobody HA4 and 712 monobody-AsLOV2 chimeras were grown in 250 mL of autoinduction media + Kan for 16 hours 713 at 30˚C. For each test, monobody HA4 (used as control) and 3 different monobody-AsLOV2 714 chimeras were tested simultaneously. Cells were then harvested by centrifugation at 7500 xg for 715 20 minutes at 4˚C in a Lynx 4000 centrifuge (SorvallÔ) and supernatant was discarded. Cell 716 pellets were resuspended in Binding buffer (Tris 100 mM pH 8.0, NaCl 150 mM, Glycerol 1% 717 and 5 mM imidazole) supplemented with 1 mM of PMSF, adding 8 mL to the cells containing

718 His6-YFP-SH2 and 3 mL for the monobody HA4 and each of the chimeras. The resuspended 719 cells were flash-frozen forming droplets directly into liquid nitrogen (LN2) and placed in small

26 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

720 (LN2-cold) grinding vials to be disrupted using a CryoMill system (Spex sample prep®) with a 721 cycle of 2 min grinding and 3 min cooling (14 times). The broken cell powder was thawed in 50 722 mL Falcon tubes at room temperature with the addition of 4 mL of Lysis buffer (Binding buffer

723 supplemented with 1 mM PMSF and 2 mg/mL of DNAse) for His6-YFP-SH2 and 2 mL of Lysis 724 buffer for monobody HA4 and each of the chimeras. Once thawed, the bacterial lysates were 725 centrifuged at 25,000 xg in a Lynx 4000 centrifuge (SorvallÔ) for 30 minutes at 4˚C and the 726 supernatant (clarified lysate) was transferred to 15 mL conical tubes. To enhance AsLOV2 727 activity, we added flavin mononucleotide (FMN) to a final concentration of 0.25 mg/ml, to each 728 of the chimera-containing lysate and then incubated for 15 minutes at 4˚C by vertical rotation in

729 a tube Revolver/Rotator at 50 rpm (Thermo ScientificÔ). The His6-YFP-SH2 supernatant was 730 then mixed with 4.5 mL of a 50% suspension of Co-charged agarose resin (Talon®), previously 731 equilibrated in Binding buffer, and incubated at 4˚C, rocking for 45 min. The suspension was 732 sedimented by gravity and the supernatant discarded. The beads were then thoroughly washed 733 with Binding buffer. with approximately 50 times the resin or column volume. The beads (now

734 with His6-YFP-SH2 bound) were then finally resuspended in Binding buffer to a total volume of 735 13.5 mL. The bead suspension was equally divided into 9 conical tubes of 15 mL (having each 736 1.5 mL of the bead suspension). Four of these tubes were used for experiments under blue light 737 with LED panels (450 nM) while the other four were used for experiments in the dark (wrapped 738 in aluminum foil). Experiments were performed in a dark room and red light was used

739 occasionally for visualization purposes. The remaining tube was left as a control for His6-YFP- 740 SH2 binding. Samples of 2.5 mL of bacterial clarified lysates containing either the monobody 741 HA4 supernatant or each one of the chimeras were added to separate tubes of resin. The mixtures 742 were then incubated for 45 minutes (at 4˚C and constant vertical rotation of 20 rpm) under blue 743 light or dark conditions. Tubes were then allowed to settle at 4˚C under the same light conditions 744 in which they were incubated, for 15-20 minutes. After carefully discarding the supernatants, the 745 beads were washed 5 to 6 times; each time with 10 mL of Binding buffer and rotating at 20 rpm 746 for 15 min. After each wash, the mixtures were allowed to settle at 4˚C for 15-20 minutes under 747 the same light conditions as they were incubated and the supernatants were again discarded. This 748 step was repeated until beads were washed with approximately 40-50 resin-volumes. During the 749 course of the experiments, light conditions within the room were carefully held constant and red 750 light was for visualization purposes used only when needed, specially to minimize any blue light

27 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

751 exposure of the dark samples when opening the wrapped tubes to exchange buffer (washing, 752 incubation and elution). After the last wash, proteins were eluted with 2.5 mL of Elution buffer 753 (Tris 100 mM pH 8.0, 150 mM NaCl, 1% Glycerol, 500 mM Imidazole) and then equal volumes 754 of elution samples (dark and light) for each chimera (and control) were loaded onto 12% 755 polyacrylamide gels and resolved with SDS-PAGE. 756 757 Purification of Monobody HA4, OptoMB (and all its variants), and YFP-SH2 758 HA4, monobody-AsLOV2 chimeras (OptoMB and variants), and YFP-SH2 constructs were 759 purified using N-terminal 6x-histidine tag fusions and metal affinity chromatography. Chimeras 760 were protected from excessive ambient light exposure to prevent potential protein destabilization 761 and improve yields. This included covering the shakers with black blankets during expression, 762 wrapping culture flasks and tubes (containing crude or purified proteins) with thick aluminum 763 foil and performing the chromatography in the dark or with red light (when needed). OptoMB 764 (and all its variants) were expressed at 18°C for three days in 1 L or 2 L of Autoinduction 765 media67 (plus Kan 50 µg/mL). HA4 and YFP-SH2 were expressed in 1 L or 2 L of Autoinduction 766 media (plus Kan 50 µg/mL) for 16 h at 30°C. Cells were then harvested by centrifugation at 7500 767 xg for 20 minutes at 4˚C in a Lynx 6000 centrifuge (SorvallÔ) and supernatant was discarded. 768 Each cell pellet was resuspended in between 8 to 12 mL of Binding buffer and frozen droplets 769 were prepared as described above and immediately transferred to a large (LN2-cold) grinding 770 tube. Cryogenic grinding was performed as described above. Broken cells (frozen powder) were 771 thawed in 50 mL Falcon tubes at room temperature with the addition of Lysis buffer up to 5% of 772 the initial cell culture volume. After clarifying the bacterial lysates by centrifugation as described 773 above, these were loaded onto columns of 2 to 5 mL (50% suspension) of Co-charged resin 774 (Talon®) previously equilibrated with Binding buffer (Tris 100 mM pH 8.0, NaCl 150 mM, 775 Glycerol 1% and 5 mM Imidazole). Columns were washed with 40-50 column volumes of 776 Binding buffer and proteins eluted with Elution buffer (Binding buffer supplemented with 250 777 mM of imidazole). Proteins were then run through size exclusion chromatography (SEC) using a 778 Hiprep™ HR 16/60 Sephacryl™ 200 with Buffer A (Tris 50 mM pH 8.0 and NaCl 150 mM), in 779 an FPLC (ÄKTA pure from GE® Healthcare). The aliquots enriched with the target proteins 780 were concentrated by centrifugation (in cycles of 10 min) at maximum speed in a Sorvall Legend 781 XTR Benchtop centrifuge (Thermo ScientificÔ) using 15 mL centricons® (Millipore) with a

28 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

782 cutoff selected according to the molecular size of each protein, until concentrations between 3 to 783 8 mg/mL were reached. 784 785 Size-exclusion chromatography to characterize OptoMB-SH2 interaction in solution 786 Purified monobody HA4 (75 µL from a 202 µM sample) or OptoMB (wt AsLOV2) (300 µL 787 from a 40.7 µM sample) dissolved in Buffer A, were mixed with YFP-SH2 (200 µL from a 58.8 788 µM sample in Buffer A) in approximately 1.2:1 molar ratio of binder to target. Each mixture was 789 incubated for 10 min in either dark or blue light (450 nm) before loading the full volume (275 or 790 500 µl) onto a Superdex™ 200 16/300 column (GE® Healthcare), which was then run at 1 791 mL/min with Buffer A at 4˚C (using an ÄKTA pure from GE® Healthcare). To test different 792 light conditions, the column was either illuminated with wrapped blue LED strips (450 nm) or 793 covered with aluminum foil for the total duration of the filtration (Supplementary Fig. 2c). For 794 experiments in the dark, we also minimized light sources in the room and covered the 795 chromatography cabinet with a black blanket. The SEC was monitored by UV absorbance at 280 796 nm. 797 798 Intracellular imaging of HEK293T cells 799 For live cell imaging, we used 0.17 mm glass-bottomed black walled 96-well plates (In Vitro 800 Scientific). Glass was first treated with 10 µg/mL of fibronectin in PBS for 20 min. HEK293T 801 cells expressing both SH2-mCherry-CAAX and either the monobody HA4 or OptoMB were then 802 plated and allowed to adhere onto the plate before imaging. Mineral oil (50 µL) was added on 803 top of each well with cells prior to imaging, to limit media evaporation. The mammalian cells

804 were kept at 37˚C with 5% CO2 for the duration of the imaging experiments. The irFP and 805 mCherry fluorescence were imaged using a Nikon Eclipse Ti microscope with a Prior linear 806 motorized stage, a Yokogawa CSU-X1 spinning disk, an Agilent laser line module containing 807 405, 488, 561 and 650 nm lasers, an iXon DU897 EMCCD camera, and a 40X oil immersion 808 objective lens. An LED light source was used for photoexcitation with blue light (450 nm), 809 which was delivered through a Polygon400 digital micro-mirror device (DMD; Mightex 810 Systems). 811 812 Imaging of coated Agarose Beads

29 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

813 Approximately 200 µL of Ni-NTA agarose slurry (50% suspension) (Qiagen) equilibrated in 814 Buffer A were mixed with 500 µL of 100 µM of either monobody HA4 or OptoMB (AsLOV2 815 V416L variant) in an Eppendorf tube (covered with aluminum foil) and incubated by vertical 816 rotation (at 20 rpm for 20 min) at room temperature to allow binding through the 6x-histidine tag 817 until saturation. The excess protein was then washed twice with 1 mL of Buffer A, centrifuging 818 the beads at low speed for 1 min (1000 rpm in a benchtop centrifuge) each time; finally 819 discarding the excess of supernatant after the second wash. Then, 50 µL of a purified YFP-SH2 820 solution at 2 µM (with the 6x-histidine tag cleaved off) was added onto 0.17 mm glass-bottomed 821 black walled 96-well plate (In Vitro Scientific) followed by 2 µL of the washed resin with the 822 beads labeled with OptoMonobody (or monobody HA4 as control). The mixture was equilibrated 823 for at least 1 hour at room temperature and up to overnight at 4°C prior to imaging, performed at 824 room temperature. The same microscope setup as described for the cells imaging was used, 825 except for the objective (20X in this case), to follow YFP fluorescence over time on the surface 826 of the bead. For the spatial control of the OptoMB-SH2 interaction on beads the same setup was 827 used. Two beads in an area of around 200 x 250 µm were imaged at the same time applying a 828 light (450 nm) mask, which uses a circular ROI with a radius of 70 µm to cover and illuminate 829 only one bead. The YFP fluorescence was recorded over time (for a total of 1h) for both, the 830 illuminated and the unilluminated bead. Quantification was performed by measuring the change 831 in YFP fluorescence intensity over time in a defined region on the surface of the bead (using 832 ImageJ68) and subtracting the background. 833 834 Calculation of binding kinetics by Bio-layer interferometry

835 Measurements of the binding (kon) and unbinding (koff) rate constants, as well as the dissociation

836 (or affinity) constant (Kd) for HA4 monobody and OptoMB (including variants V416L and 837 V416I-G528A-N538E) were performed on Octet RED96e instruments (ForteBio). Ni-NTA 838 sensors (ForteBio) were first equilibrated using Buffer A for 10 min prior the measurement. A 839 volume of 200 µL of Buffer A or protein solutions (previously dialyzed with Buffer A when 840 needed) were added to clear 96-well plates. During the experimental run, the Ni-NTA sensors 841 were first immersed in Buffer A to record the baseline. Protein binders were then loaded by 842 switching to wells with solutions of 6x-histidine tagged HA4 monobody or OptoMB variants 843 (with concentrations between 100 µg/mL and up to 1 mg/mL) until values of ~ 4 nm were

30 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

844 reached (avoiding saturation of the sensors). The sensors were then transferred back into Buffer

845 A to remove unbound protein. To measure the binding rate constant (kon) the sensors with bound 846 monobody HA4 or OptoMB variants were subsequently shifted to wells containing various 847 concentrations of YFP-SH2 (at concentrations indicated Fig 2d-e and Supplementary Fig. 2e-f).

848 To measure the unbinding rate constant (koff), the sensors were then moved to wells containing 849 Buffer A to trigger dissociation of YFP-SH2. To measure binding kinetics of the light state, the 850 lid of the Octet remained open during the measurement and a blue LED panel was held above the 851 96-well plate, maintaining constant blue illumination for the duration of the experiment. For the 852 OptoMB variant with the AsLOV2 mutations V416I-G528A-N538E, lit states were sufficiently 853 long-lived to remain fully activated in response to pre-illumination with a blue light panel and 854 the continuous illumination of the Octet sensors. The raw binding and unbinding data were 855 simultaneously fit to models of the binding and unbinding reactions: 856

ii-(kkton [SH2] +off ) i-kt 857 yt( ) =éù a1- ei + beleak bind ëûêúon ( ) on 858

-kt 859 ytaebeiii=+éùoff -ktleak unbind ( ) ëûoff off 860 861 This model incorporates the following dependent and independent variables:

i th 862 ytbind ( ) refers to the i binding curve

i th 863 ytunbind ( ) the i unbinding curve

th 864 [SH2]i refers to the concentration of YFP-SH2 used for the i binding curve 865 t is the time elapsed since the start of the binding/unbinding phase. 866 It also includes the following parameters:

867 kon is the on-rate (enforced to be identical across all binding and unbinding curves)

868 koff is the off-rate (enforced to be identical across all binding and unbinding curves)

869 kleak represents the slow unbinding of His-tagged OptoMB from the probe, leading to a gradual 870 decay of signal throughout the entire experiment.

i th 871 aon is the total change in signal due to SH2 binding for the i curve

i th 872 bon is the signal baseline during the binding phase for the i curve

31 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

i th 873 aoff is the total change in signal due to SH2 unbinding for the i curve

i th 874 boff is the signal baseline during the unbinding phase for the i curve 875 The model thus contains 4*n + 3 parameters, where n is the number of distinct SH2 876 concentrations tested. Fits were performed using nonlinear gradient descent using the MATLAB 877 fmincon function. Code is available as described in Ref. 45. 878 879 Light- controlled affinity chromatography (LCAC) of SH2-tagged proteins using cobalt- 880 immobilized OptoMB 881 To prepare our OptoMB resin and columns, we incubated the purified variant

882 OptoMBV416I_G528A_N538E (having a SUMO tag in the N-term to boost expression, Supplementary 883 Table 4) with Co-charged resin (Talon®) (50% suspension) previously equilibrated in Buffer A, 884 in order to produce 2.5 mL of resin with approximately 8 mg of OptoMB bound per mL. The 885 Resin was rotated in 50 mL conical tubes (covered with aluminum foil) at room temperature for 886 20 min and then poured in 20 mL BioRad glass Econo-Columns and washed with Buffer A (50 887 column volumes). The use of imidazole in the Buffer A was avoided because it interferes with 888 the photocycle of the AsLOV2 domain, enhancing the rate of recovery of the dark state thus 889 reducing the lifetime of the lit state69, which would reduce the efficiency of light elution. 890 Production of the OptoMB resin (and columns) and the LCAC procedure were carried out at 891 room temperature within a carefully controlled dark room to avoid blue light contamination, 892 using red light from LED panels occasionally for visualization purposes. Approximately 10 to 12 893 mL of clarified bacterial lysates (obtained as described before from 1L cultures) containing YFP- 894 SH2 or SH2-PDC1 (without his-tags) were incubated with 2.5 mL of OptoMB resin in Buffer A, 895 which was enough to saturate the OptoMB resin with binding targets. The mixture was then 896 vertically rotated gently for 45 min in 15 ml conical tubes. The OptoMB resin was then poured 897 into 10 mL BioRad glass Econo-Columns or left to settle to discard the supernatant by pipetting 898 (experiments in batch). Approximately 80 column volumes of Buffer A were used to wash the 899 OptoMB resin. When purifying in batch, 40 mL of Buffer A were added to the resin at a time in 900 50 mL conical tubes; after rotating for 5 min, the resin was left to settle to remove the 901 supernatant by pipetting. This cycle was repeated until reaching 80 resin volumes. After 902 washing, the resin was poured back into a 15 mL conical tube. When purifying in batch, for each

32 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

903 light-elution step we added 2-times the resin volume of Buffer A and incubated with gentle 904 rocking in front of a blue light (450 nm) LED panel (approximately 20 cm away) for 10 to 15 905 min; then letting the resin settle while illuminated and recovering the light-eluted aliquot by 906 pipetting. When purifying in columns, 2 column bed volumes of Buffer A were poured at a time 907 while the column was exposed to blue light from three LED panels in a triangular conformation 908 (surrounding the column), and collected by gravity. For both batch and column purifications, the 909 elution steps were repeated until no more protein eluted from the resin (usually 4 to 5 times). In 910 both methods, batch or column, after light-elution we added 4 times the resin volume of Buffer A 911 containing 250 mM of imidazole to elute the proteins still bound to the resin after light-elution, 912 needed to calculate yields and resin capacity. Samples from the different purification steps were 913 resolved with SDS-PAGE, using 12% SDS-polyacrylamide gels. 914 915 Light-controlled affinity chromatography (LCAC) of SH2-tagged proteins using covalently 916 coupled OptoMB 917 A total of 1-1.5g (dry weight) of CNBr-activated SepharoseTM 4B (GE® Healthcare) was

918 resuspended in Conjugation buffer (NaCHO3 0.1M pH 8.3 and 500 mM NaCl) according to the 919 manufacturer specifications. The excess of Conjugation buffer was discarded and the hydrated

920 resin mixed with purified OptoMBV416I_G528A_N538E (containing an N-terminal SUMO tag to boost 921 expression, Supplementary Table 4), previously dialyzed against Conjugation buffer by 922 centrifugation (using 15 mL centricons® as describe above). For each preparation, 5 mL (50% 923 suspension) of resin was conjugated with OptoMB at approximately 8 mg of OptoMB per mL of 924 resin. The manufacturer’s protocol was followed with minor modifications, keeping all protein 925 immobilization steps (conjugation, blockage and acid/base washing cycles) in the dark by 926 covering the tubes with thick aluminum foil. The acid/base washing cycles were reduced to one 927 cycle, as the three washes recommended by the manufacturer results in loss of the FMN, the 928 AsLOV2 co-factor (as evidenced by the loss of the protein’s yellow coloration immediately after 929 the second wash). Once OptoMB was conjugated, the resin was washed and equilibrated with 930 Buffer AS (Buffer A with salt concentration increased to 300 mM NaCl). The OptoMB- 931 conjugated resins were used immediately or stored at 4˚C (protected from light with foil), where 932 they are stable for up to at least a week (yielding same results as fresh resin). To purify YFP-SH2 933 or SH2-PDC1, approximately 10 to 15 mL of the clarified bacterial lysates obtained from 1 L

33 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

934 cultures (as described above) were incubated in batch with 5 mL (50% suspension) of the 935 OptoMB-conjugated sepharose resin, which was enough to saturate the OptoMB resin with 936 binding targets. The same protocol for LCAC purification with OptoMB-immobilized in Co- 937 charged (Talon®) in batch described above was followed with CNBr-OptoMB resin, except 938 Buffer AS was used instead of Buffer A. To calculate the yields and resin capacity, after the final 939 light-elution step, 200 µl of the 50% resin suspension were resuspended in 100 µl of SDS-PAGE 940 sample-loading buffer and incubated at 100˚C for 10 min in a heat block, before loading onto the 941 gel. Samples from all purification steps were resolved on a 12% SDS-polyacrylamide gels. 942 943 Calculations of Protein Purity, Yield and Resin capacity 944 Protein purity was estimated using densitometric analysis in the FIJI implementation of 945 ImageJ68. The software analyzes individual bands from a scanned gel, returning the integrated 946 intensity of each band. For each aliquot of purified protein (eluted with light), we selected the 947 band representing the SH2-tagged protein, and used their integrated intensities to calculate the 948 fraction of the purified target versus the overall total of protein present in that fraction. 949 950 The overall yield (y) represents the total amount (mg) of purified protein that eluted with light 951 (TPE) as a percentage of the total amount (mg) of protein that was initially bound to the resin 952 (TPB). TPE was calculated by multiplying the concentration (mg/mL) of all the light-eluted 953 fractions ([TPE]) by their total volume (mL) (TPEvol). The concentrations of the fractions eluted 954 with light were calculated using Absorbance at 280 nm with a spectrometer (Biospectrometer, 955 Eppendorf). TPB was equal to TPE plus the total residual SH2-tagged protein remaining on the 956 column after light-elution (TPR). To calculate the TPR, we recovered all residual protein using 957 imidazole elution (of Co-Agarose-OptoMB) or heat treatment (of CNBr-OptoMB) of the beads 958 after light-elution. Concentrations of recovered residual protein were calculated by running each 959 sample on a 12% polyacrylamide gel alongside standards of the same SH2-tagged protein of 960 known concentrations. Band intensities were then computed by densitometry analysis in FIJI as 961 above. The amount of target protein still bound to the column was thus equal to the concentration 962 retrieved in this final elution [TRP] multiplied by its total elution volume TRPvol. Finally, the

963 resin capacity CR was calculated as TPB divided by the total volume of the resin used VR. 964

34 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

�� × 100 965 � = �� 966 �� = [��] × �� 967 �� = {[��] × ��} + �� 968 � = � 969 Where 970 y is the yield of purified protein obtained with the resin 971 TPE is the total amount of protein eluted with light 972 [TPE] is the concentration of all the fractions that eluted with light 973 TPEvol is the total volume of the light-elution fraction 974 TPB is the total amount of protein initially bound to the resin 975 [TRP] is the concentration of the total residual protein fraction that was still bound to the resin 976 after the elution with light 977 TRPvol is the total volume of the fraction containing the total residual protein that was still 978 bound to the resin after the elution with light, and was recovered by imidazole or heat treatment 979 of the beads.

980 CR is the protein binding capacity of the resin

981 VR is the volume of resin used in the experiment 982 983 Data Availability 984 The authors declare that all data supporting the findings of this study are available within the 985 paper (and its supplementary information files), but original data that supports the findings are 986 available from the corresponding authors upon reasonable request. 987 988 Author contributions 989 Conceptualization, C.C.L., E.M.Z., A.A.G., J.E.T. and J.L.A.; Methodology, C.C.L., E.M.Z., 990 A.A.G., J.E.T. and J.L.A.; Investigation, C.C.L., E.M.Z, A.A.G, and N.A.; Writing – Original 991 Draft, C.C.L., E.M.Z. and J.L.A.; Review & Editing, all authors; Funding Acquisition, J.E.T. and 992 J.L.A.; Supervision, J.E.T. and J.L.A. 993

35 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.08.831909; this version posted March 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

994 Acknowledgements 995 We thank all members of Avalos and Toettcher labs for helpful comments. We also thank the 996 Biophysics Core Facility and especially Venu Vandavasi for help with the bio-layer 997 interferometry measurements. J.E.T. was supported by NIH grant DP2EB024247 and J.L.A was 998 supported by the U.S. Department of Energy, Office of Science, Office of Biological and 999 Environmental Research Award Number DE-SC0019363, the NSF CAREER Award CBET- 1000 1751840, The Pew Charitable Trusts, The Eric and Wendy Schmidt Transformative Technology 1001 Fund Award, and the Camille Dreyfus Teacher-Scholar Award. 1002